diff --git a/cluster.go b/cluster.go
index 431bd7c61..c4a295072 100644
--- a/cluster.go
+++ b/cluster.go
@@ -50,9 +50,8 @@ const (
 	ClusterStateResizing = disco.ClusterStateResizing
 	ClusterStateDown     = disco.ClusterStateDown
 
-	// NodeState represents the state of a node during startup.
-	nodeStateReady = "READY"
-	nodeStateDown  = "DOWN"
+	// nodeStateDown represents the state of a node which is unavailable.
+	nodeStateDown = "DOWN"
 
 	// resizeJob states.
 	resizeJobStateRunning = "RUNNING"
@@ -213,24 +212,6 @@ func (c *cluster) unprotectedIsCoordinator() bool {
 	return snap.PrimaryFieldTranslationNode().ID == c.Node.ID
 }
 
-// unprotectedSendSync is used in place of c.broadcaster.SendSync (which is
-// Server.SendSync) because Server.SendSync needs to obtain a cluster lock to
-// get the list of nodes. TODO: the reference loop from
-// Server->cluster->broadcaster(Server) will likely continue to cause confusion
-// and should be refactored.
-func (c *cluster) unprotectedSendSync(m Message) error {
-	var eg errgroup.Group
-	for _, node := range c.noder.Nodes() {
-		node := node
-		// Don't send to myself.
-		if node.ID == c.Node.ID {
-			continue
-		}
-		eg.Go(func() error { return c.broadcaster.SendTo(node, m) })
-	}
-	return eg.Wait()
-}
-
 // addNode adds a node to the Cluster and updates and saves the
 // new topology. unprotected.
 func (c *cluster) addNode(node *topology.Node) error {
diff --git a/holder.go b/holder.go
index 48ca449f1..d697b3ab4 100644
--- a/holder.go
+++ b/holder.go
@@ -1823,6 +1823,11 @@ type holderCleaner struct {
 	Closing <-chan struct{}
 }
 
+// TODO: this is here to satisfy the linter since holderCleaner was removed from
+// the gossip implementation of removeNode. But presumably we will use it once
+// we have ported over the etcd implementation.
+var _ holderCleaner
+
 // IsClosing returns true if the cleaner has been marked to close.
 func (c *holderCleaner) IsClosing() bool {
 	select {